Non-canonical functions of UHRF1 maintain DNA methylation homeostasis in cancer cells

Abstract

DNA methylation is an essential epigenetic chromatin modification, and its maintenance in mammals requires the protein UHRF1. It is yet unclear if UHRF1 functions solely by stimulating DNA methylation maintenance by DNMT1, or if it has important additional functions. Using degron alleles, we show that UHRF1 depletion causes a much greater loss of DNA methylation than DNMT1 depletion. This is not caused by passive demethylation as UHRF1-depleted cells proliferate more slowly than DNMT1-depleted cells. Instead, bioinformatics, proteomics and genetics experiments establish that UHRF1, besides activating DNMT1, interacts with DNMT3A and DNMT3B and promotes their activity. In addition, we show that UHRF1 antagonizes active DNA demethylation by TET2. Therefore, UHRF1 has non-canonical roles that contribute importantly to DNA methylation homeostasis; these findings have practical implications for epigenetics in health and disease.

Fig. 1. Establishment and validation of endogenous AID-tagged UHRF1 and/or DNMT1 HCT116 cells.

A Schematic of the CRISPR/Cas9 genome editing strategy to endogenously tag UHRF1 with mAID/mClover and DNMT1 with mAID/mRuby2. B Order of events for the generation of the different cell lines. C Immunoblot images for validation of endogenous AID-tagged UHRF1 and/or DNMT1 HCT116 cells. Experiments in each panel were performed at least three times, and the representative results are shown. D Representative fluorescence images on UHRF1-AID/DNMT1-AID HCT116 cells showing that tagged UHRF1 and DNMT1 co-localize. E Quantification of the DNA methylation level in each HCT116 cell line with LUMA, LC-MS/MS, or WGBS. The p value is calculated with one-way ANOVA and Tukey’s HSD test (*p < 0.05). Data are presented as mean values +/− SEM from biological triplicates. Source data are provided as a Source Data file.

Fig. 2. The depletion of UHRF1 and DNMT1 is efficient, negatively affects growth, and can be rescued genetically.

A Immunoblot of HCT116 cells following treatment with Auxin (IAA) at the indicated time points (hours) and before treatment (NT). Experiments in each panel were performed at least three times, and the representative results are shown. B Cell proliferation of the HCT116 derivatives in the continuous presence of auxin for the indicated durations (Incucyte videomicroscopy). Error bars represent the SEM of biological triplicates. C Cell proliferation of the HCT116 derivatives in the continuous presence of auxin for the indicated durations (cell counting). The error bars represent the SEM of biological triplicates. D Schematic of the rescue experiments. E UHRF1 domain map showing the mutants studied (left panel) and corresponding cell proliferation analysis (Cell count, right panel). Error bars represent the SEM of three independent experiments. F Same as (E), but for DNMT1. Source data are provided as a Source data file.

Fig. 3. UHRF1-depleted cells show more severe DNA hypomethylation than DNMT1-depleted cells.

A Global DNA methylation analysis in the indicated HCT116 derivatives after auxin treatment for the indicated duration (LUMA). Error bars represent the SEM of biological triplicates. B As in (A), but the quantitation of 5-mC was done by LC-MS/MS. C As in (A), but the quantitation of DNA methylation was done by WGBS. D LUMA analysis on cells treated with auxin or 5-aza. U: UHRF1-AID; D: DNMT1-AID; UD: UHRF1-AID/DNMT1-AID. E Western blotting on 5-Aza-treated cells. Experiments in each panel were performed two times, and the representative results are shown. Source data are provided as a Source data file.

Fig. 4. Greater loss of DNA methylation upon UHRF1 depletion than upon DNMT1 depletion; UHRF1 regulates DNA methylation at DNMT1, DNMT3A and DNMT3B sites.

A Volcano plot of differentially methylated regions (DMRs, 1 kb bins) after 4 days of depletion of UHRF1 and/or DNMT1. Blue dots: hypomethylated regions (>25% loss of methylation, q-value < 0.01), red dots: hypermethylated regions (>25% gain of methylation, q-value < 0.01). Gray dots: no significant change. The p value is corrected to q-value using sliding linear model (SLIM). B Venn diagram of the hypomethylated regions in the indicated cell lines, 4 days after depletion of the proteins. C Workflow used to quantitatively compare WGBS methylation values to the in vitro preferences of DNMT1, DNMT3A and DNMT3B. D Higher levels of H3K36me3 correlate with larger losses of DNA methylation in CpG islands. The CGIs were ranked by H3K36me3 level in HCT116 cells and split into 10 equally sized bins. DNA methylation differences were estimated from biological triplicates. Lines = median; box = 25th–75th percentile; whiskers = 1.5 × interquartile range from box. The p value is from Spearman’s correlation tests. Source data are provided as a Source data file.

Fig. 5. Physical and functional interaction between UHRF1, DNMT3A, and DNMT3B.

A Western blotting after the indicated co-immunoprecipitation experiments. hUHRF1: Full-length protein. The other names indicate isolated domains, as depicted in Fig. 2E. B Same as in (A), except we used truncated constructs in which the indicated domains were deleted from the full-length protein. C Same as in (A), except we used a full-length UHRF1 protein with a point mutation in the Tandem Tudor Domain (Y188A/Y191A). D Western blotting showing abundance of the indicated proteins in total cell extracts. E Quantitation of the loss of DNA methylation in the indicated cell lines after 8 days of protein depletion, by LC-MS/MS. The p value is calculated with one-way ANOVA and Tukey’s HSD test (N.S. p > 0.05, **p < 0.01). Data are presented as mean values +/− SEM from biological triplicates. A–D Experiments in each panel were performed at least two times, and the representative results are shown. Source data are provided as a Source data file.

Fig. 6. UHRF1 protects against active demethylation by TET2.

A Workflow used to quantitatively compare WGBS methylation values to the in vitro preferences of TET1 and TET2. B RT-qPCR analysis for validation of TET2 knockdown for HCT116 UHRF1 and/or DNMT1-AID cell lines. C Global DNA methylation analysis (LUMA) for HCT116 UHRF1 and/or DNMT1-AID cell lines combined with TET2 knockdown. D LC-MS/MS analysis on the indicated samples. Error bars represent the SEM of three independent experiments. The p value is calculated using two-sided Student’s t test (N.S. p > 0.05, ***p < 0.001). Source data are provided as a Source data file.

Fig. 7. A revised and expanded model for UHRF1 functions in DNA methylation homeostasis.

Left panel: the canonical function of UHRF1 is to promote DNA methylation maintenance by DNMT1. Right panel: we demonstrate non-canonical functions of UHRF1 that are independent of DNMT1 but contribute to DNA methylation homeostasis.